Research On Short Term Traffic Flow Forecasting Based On Echo State Network

With the continuous development of urban economy and society, urban traffic volume is increasing and the traffic problem is becoming more and more serious. The most serious problem is traffic congestion. Intelligent transportation system(ITS) is an effective and new method to alleviate urban traffic pressure and reduce environmental pollution. The key problem of traffic signal control and traffic flow guidance system is real-time and accurate traffic flow forecasting. The traffic problem becomes more and more serious with the increase of the traffic volume in the city. The prediction accuracy is directly related to the traffic control and traffic guidance.Through the characteristics of city road traffic flow nonlinearity, uncertainty and the previous methods of traffic flow prediction, the echo state network is used in traffic flow forecasting in this paper, and some methods are improved and optimized.The main research works of this paper are as follows:(1) In this paper, two important parameters of the reconstructed phase space are obtained by using the C-C method, and the phase space reconstruction of traffic flow time series is reconstructed. In this basis, the Lyapunov index is obtained by using the minimum number method, and the chaotic characteristic of traffic path is determined by the index.(2) Through the principle and method of echo state network algorithm, some defects in the topology of the echo state network reserve pool are proposed, and the topology structure connection mode is improved on the basis of the proposed method,which reduces the coupling between neurons in the reserve pool and enriches its dynamic characteristics. Finally, the improved echo state network with topology structure is used to predict the experiment and obtain better results.(3) Based on the global search optimization of genetic algorithm, the echo state network is combined with the genetic algorithm, which avoids the possibility that echo state network fall into local optimal solution and improves accuracy of traffic flow prediction.